OpenAI is awarding $1 million in grants to 14 projects led by independent organizations, alongside up to $1 million in API credits, to explore how societies can respond to increasingly capable AI. The initiative follows OpenAI’s April 2026 Industrial Policy for the Intelligence Age paper and is intended to move policy ideas from theory into practical research, prototypes and frameworks. More than 400 organizations and individuals submitted proposals.
One major theme is broadening economic opportunity. Funded projects will examine AI's effects on employment and skills, how productivity gains are distributed across European workers and regions, and whether people should have a broader “Right to AI.” Other work will explore AI's impact on taxation, social benefits and economic preparedness. The programme also includes research into how states can expand electricity generation and transmission to meet growing data-center demand.
The second theme is societal resilience and AI safety. Projects will investigate recursively self-improving AI, information-sharing around AI-enabled biological risks, frontier-AI safety training for national-security professionals, privacy-preserving AI agents for government policy simulations and AI-assisted evaluation of democratic accountability. A Brazilian public-hospital project will also test an AI clinical-information infrastructure designed to improve workflows while preserving physician judgment and human oversight.
The initiative is notable because OpenAI explicitly argues that technology companies should not make these policy decisions alone. The selected projects will run for six months, with results expected in 2027, and are intended to give governments and democratic institutions evidence they can use when designing policy. The larger message is that the AI transition will require more than better models: it will require new approaches to work, ownership, taxation, energy, healthcare, safety, privacy and democratic oversight. In that sense, OpenAI is trying to widen the policy conversation beyond AI developers themselves—while also putting its own ideas into practice and inviting independent institutions to test and challenge them.
